Recent Advances in Active Noise Control Inside Automobile ...€¦ · In an attempt to resolve the above problem, Active Noise Control (ANC) methods were developed where secondary

1

Recent Advances in Active Noise Control

Inside Automobile CabinsPrasanga N. Samarasinghe*, Member, IEEE, Wen Zhang, Member, IEEE

and Thushara D. Abhayapala, Senior Member, IEEE,

I. INTRODUCTION

Minimization of interior cabin noise has been a key topic of research in the automobile industry for the

last two decades. This problem was initially approached via passive noise cancellation methods, where

physical treatments such as structural damping and acoustic absorption were used. However, with vehicle

manufacturers striving for more economical and light weight designs, the resulting car interiors invariably

became noisier due to the increased structural vibrations. These noise fields are generally dominanted by

low frequencies (i.e, 0− 500 Hz) [1], [2], hence the conventional passive noise cancellation approaches

are less effective. In an attempt to resolve the above problem, Active Noise Control (ANC) methods

were developed where secondary sources were proposed to attenuate the noise inside the cabin. With

modern in-car entertainment systems providing 4 − 6 built-in loudspeakers, the addition of an Active

Noise Control system comes at a relatively low additional cost.

In practice, in-car ANC is achieved by producing a secondary signal(s) that cancels the noise generated

by the noise source(s). The residual difference between these two components is measured using a

microphone(s) placed inside the cabin, and is minimized using a feed-forward/feed-back control system

[3]. Feed-forward systems use a time-advanced “reference signal(s)" correlated with the noise signal to

attenuate the primary noise field, whereas feed-back systems tend to attenuate the overall measured noise.

The basic concept behind ANC inside vehicles is described in Fig. 1, which shows how a secondary

soundfield cancels out the undesired noise field utilizing an adaptive controller. Since the noise observed

inside vehicle cabins is often random and time varying, the aforementioned control systems are required to

be adaptive. While the theory and concept behind ANC systems are quite straightforward, their practical

implementation and performance are often hindered by factors such as the noise field complexity inside

the car geometry, cost, adaptive system convergence time, system stability, non-causality and poor spatial

coverage. Over the last 30 years researchers and car manufacturers have done extensive amounts of

experimental research to overcome these limitations [1], [4]–[6].

August 19, 2016 DRAFT

2

Built-in audio

system

Microphones detecting the residual error

The rotational frequency of the engine

Speakers at the front

Speakers at the back

Adaptive Controller

Audio System

Engine

Fig. 1. Active Noise Control inside an automobile cabin

In this article, we review ANC techniques for noise cancellation inside automobiles. We explain the

different types of noise fields present inside vehicles, the theoretical basics of ANC techniques and their

applicability in canceling the aforementioned noise fields. We focus on recent advances made in vehicle

ANC during the past 10 − 15 years including commercial developments available in mass production

vehicles. We aim to show that in-car ANC is an exciting field of research with the potential to substantially

improve the passenger experience in acoustically challenging environments.

II. NOISE SOURCE AND NOISE FIELD INSIDE VEHICLES

Different kinds of noise sources exist in automobiles, such as engine noise, road-tyre noise, and wind

noise, with their own distinct acoustic properties. The vehicle compartment can be considered a very

small room and depending on the frequency, the acoustics inside the car exhibit completely different

physical behaviors. This section reviews the attributes of typical noise sources, and noise models, which

effectively describe the noise fields inside automobiles.

A. Attributes of Noise Sources

1) Engine Noise: Most road vehicles with four or more wheels employ a reciprocating, four stroke,

internal combustion engine [7]. The noise produced by the internal combustion engine is dominated by

two processes, the piston crank mechanism and the combustion process. The piston crank mechanism,

such as the movement of pistons and their lines, generate an impulsive noise with a flat spectrum; the

combustion process on the other hand produces a tonal noise, which is directly related to the rotational

speed of the engine. ANC strategies are generally more effective in controlling combustion noise processes

due to its predictive nature. Based on in-car noise measurements, there exist two simple relationships


3

between the engine type and the resulting combustion noise. Given the engine size expressed in terms of

the number of cylinders and their individual capacity in liters, the average noise level can be empirically

estimated using the following equation [6]

Noise Power Level = 10 log10(no. of cylinders) + 23 log10(cylinder capacity).

With the knowledge of the engine rotation speed, the fundamental noise frequency/firing frequency/dominant

engine order is

f0 =rotation speed

2× 60× no. of cylinders.

Derivation of the dominant engine order at any given rotation speed (rpm) is straightforward. First, the

rpm is converted to Hertz by a multiplication of 1/60 (e.g., an engine spinning at 1800 rpm can be

said to be running at 30 Hz). Second, since a four-stroke engine fires each cylinder only once every

two crank revolutions, the rotations per second is multiplied by half the number of cylinders (e.g., for

a 6 cylinder engine spinning at 1800 rpm, the dominant engine order is at 90 Hz), which gives the

fundamental frequency. In a six-cylinder engine, it’s also called the “third engine order" because the

frequency is three times that of the engine’s rotation in Hz. While the dominant engine order defines

the engine’s distinctive sound character, its overall timbre is decided by multiple variables such as its

structure, plumbing and materials, which cause additional engine orders to become active. Therefore,

typical engine induced noise carries multiple engine orders.

In automatic transmission powertrains, the torque converter and torque converter clutch are critical

devices governing the overall power transfer efficiency. They create a one to one connection between the

output of the engine and the input of the transmission. With increasing demand for fuel economy, the

recent trend is to apply the torque converter clutch over a wider range of driving conditions. This increases

powertrain high torque fluctuation and causes noise and vibration. While there exist many passive control

solutions for this issue, active control of noise and vibration is one of the most efficient solutions because

active control avoids the addition of extra weight [8].

2) Road-tyre Noise: Road-tyre noise is produced due to the interaction between the road surface and

the tyre, which can be classified mainly into two kinds, air pumping noise and vibration induced noise

(or road booming noise) [7]. Air pumping noise is caused by the tyre/tread pumping when on a rough

road surface. Normally, this kind of noise is dominated by high frequencies and its level is dependent

on factors such as the size of the road, the size of tyre cavities, the load on the tyre, and the pressure

inside the tyre. In contrast, vibration induced noise (or road booming noise), is mainly due to the non-

uniformity of road surfaces, change of vehicle speed and irregularities in the tyre tread pattern. There are

several characteristics of vibration induced noise. First, it is generated by a combination of independent


4

vibrations of the four wheels, therefore it is hardly reduced by ANC techniques with one or two reference

sensors. Second, the spectrum and properties of vibration induced noise vary continuously with varying

road profiles and vehicle speed. Third, transfer function between wheel vibrations and road booming

noise is non-linear. The above characteristics are often unique to the vehicle of interest and therefore,

ANC of road noise is rather difficult to achieve in realistic conditions as opposed to laboratory set-ups.

3) Wind Noise: The wind noise is the predominant component of interior noise at speeds above 100

km/h [2]. It can be classified by the noise production mechanisms, such as (i) a low frequency broadband

noise due to the vehicle moving through air at mid to high speeds or turbulent air flow through holes

(e.g., vehicle windows and doors), (ii) an impulsive noise above 300 Hz due to the external varying wind

conditions, and (iii) a narrowband beating noise due to air flow over open windows or sunroofs. The

acoustic energy inside vehicle cabins due to wind noise is generally concentrated at the frequency band

50− 500 Hz.

B. Noise Fields inside Vehicles

The vehicle compartment is generally considered as a small room and has distinct acoustic properties.

Similar to room modes1 At low frequencies, the sound field is dominated by a limited amount of

acoustic modes. At high frequencies, as the number of acoustic modes increases, the sound field becomes

increasingly diffuse, making it more accurate to be modeled using statistical methods. In this region, a

typical phenomenon is the spectral coloration, where the spectral peaks and dips do not correspond to

eigen frequencies of acoustic modes but are the result of numerous overlapping resonances. Often the

Schroeder cut-off frequency [9] is used to separate the low- and high-frequency regions. This is about

300 Hz in a typical car compartment [6].

In current automotive ANC systems that are commercially available, the main objective is to globally

cancel the dominant engine order(s) inside the vehicle cabin such that all passenger seats are covered.

Global cancellation of noise requires the entire noise field inside the car to be considered. This is best done

by describing the noise field in terms of acoustic modes, which depend on the noise field’s frequency

content, and the structural-acoustic coupling2 of the enclosure. ANC control systems thus require the

speakers positioned to control the above acoustic modes and the error microphones positioned to observe

1Room modes or acoustic modes are a collection of resonances that exist in a room when the room is excited by an acoustic

source. Most rooms have their fundamental resonances in the 20 Hz to 200 Hz region, each frequency being related to one or

more of the room’s dimension’s or a divisor thereof.2When a vibrating structure is in contact with air, some of the energy from the structure escape to the air as sound. This

interaction referred to as structural-acoustic coupling.


5

the relevant acoustic modes.The effective frequency range of a given ANC system is determined by the

modal density (the average number of modes in a unit frequency interval) of the automobile enclosure and

the available number and placement of microphones and loudspeakers. This limitation is purely physical

and do not depend on the control algorithm or software used for achieving ANC.

In the automotive industry, low frequency noise fields inside vehicles are often simulated using

Computer Aided Technology (CAE) to assist production. CAE often uses the finite element method

[10] and boundary element method [11] to model the vibrations and structural-acoustic coupling inside

vehicle cabins. Since CAE is largely useful for ANC, and the current direction of commercial vehicle

development is towards shortening the development time while decreasing the prototype vehicle quantity,

there is an urgent need to improve the computer based prediction technologies at the planning stage [12].

III. ACTIVE NOISE CONTROL TECHNIQUES

Active noise control involves a system that cancels the primary (unwanted) noise based on the principle

of superposition. In the time domain, for a single channel (SISO) system,

e(n) = d(n) + y′(n) (1)

where e(n) represents the error signal and d(n) and y′(n) represent the noise and secondary soundfields

present at the error sensor respectively. Typical ANC systems require one or more loudspeakers to produce

the secondary sound field, one or more microphones to measure the residual error signal(s) present at

the observation point(s) of interest, and an adaptive control system to drive the loudspeaker(s) while

minimizing the residual error. In-car ANC systems can be broadly classified as, feed-forward systems

and feed-back systems depending on whether reference sensors are present, or broadband and narrowband

depending on the operating bandwidth. The next two sections will review each system in detail, while

providing their recent applications, mainly related to engine noise cancellation and road noise cancellation.

IV. FEED-FORWARD CONTROL SYSTEMS

Feed-forward systems use an additional sensor(s) (acoustic/mechanical/optical/electric) to measure the

primary noise field or to generate a signal related to the noise generation mechanism. This reference

signal(s) is processed by the ANC system to drive the loudspeaker(s), in order to minimize the residual

error. The performance of feed-forward ANC systems is dependent on the coherence between the reference

signal and the primary noise, thus there exists an inherent requirement for the reference sensors to be

placed close to the noise source. The performance of a feed-forward ANC system also relies on the

frequency spectrum of the primary noise. If the primary noise field is a random broadband soundfield, it


6

Unknown Plant P(z)

Adaptive Filter W(z)

LMS

S(z)

Σ d(n) x(n)

y(n) y'(n)

e(n)

Fig. 2. Block diagram of a Feed-forward ANC system

is important that the reference signal and the adaptive system continuously track it. Furthermore, if the

adaptive system’s electrical delay (due to the processing time) is larger than the acoustic delay from the

reference microphone to the canceling loudspeaker, the controller response becomes non-causal and the

system performance will be substantially degraded. However, in narrowband feed-forward systems, the

continuous tracking requirement and the causality condition are largely preserved because the reference

signal is often predictable.

A. Feed-forward Control Algorithms

1) Broadband Feed-Forward ANC: Broadband feed-forward control systems are utilized when the

primary noise field has a broadband frequency response (e.g., road noise). The system identification

framework of a basic single-input single-output (SISO) broadband feed-forward control system is illus-

trated in Figure 2. The primary path system function P (z) consists of the acoustic response from the

reference sensor location to the error sensor, and the secondary path system function S(z) consists of

a combination of (i) the electronic response between the adaptive filter W (z) and the loudspeaker, and

(ii) the acoustic response from the loudspeaker to the error microphone. If the secondary path transfer

function is not taken in to account (i.e., W (z) = P (z)), the system will become unstable, because the

error signal will not be correctly time-aligned with the reference signal. Since the primary path of a

vehicle cabin is often dynamic, the role of the adaptive filter W (z) is to continuously track the time

variations in the primary noise source (through the reference signal x(n)) and minimize the residual error

signal e(n). The most common form of adaptive filters is the transversal filter using the least-mean-square

(LMS) algorithm. From Fig. 2, the Z-transform of the error signal is

E(z) =(P (z)− S(z)W (z)

)X(z) (2)


7

P(z)

W(z)

LMS

S(z)

Σ d(n) x(n)

y(n) y'(n)

e(n)

Ŝ(z)

x'(n)

Fig. 3. Block diagram of a Feed-forward ANC system using the FXLMS algorithm.

In the ideal case, E(z) = 0 after the adaptive filter converges, which implies that the optimal filter

response is W (z) = P (z)/S(z). In order to achieve this result, the adaptive filter has to simultaneously

model P (z) and inversely model S(z). Since an inverse does not always exist for S(z), Morgan [13]

suggested a more effective approach, where an identical filter (to S(z)) was proposed to be placed along

the reference signal path to the weight update of the LMS algorithm. This modification compensates for

the secondary path effects. It is also the origin of the well known filtered-X LMS (FXLMS) algorithm, for

which the corresponding block diagram is given in Fig 3. The term S(z) in Fig 3 refers to an estimated

value of the secondary path. For a more detailed derivation of the FXLMS algorithm, the reader is

encouraged to refer to [3]. Another practical limitation that arises with feed-forward systems is the effect

of feed-back. This involves the upstream of the anti-noise output from the secondary loudspeakers to the

reference sensor, which corrupts the reference signal. The simplest approach to solving this problem is to

model the feed-back path transfer function F (z) and neutralize it with a separate feed-back cancellation

filter [3]. However, this leads to additional stability issues in practice, and incorrect modeling of the

feed-back path can lead to instability.

2) Narrowband Feed-forward ANC: Noises generated by mechanical components such as engines,

compressors, motors and fans are generally narrowband and periodic. To monitor such noise, it is sufficient

to use a non-acoustic sensor (e.g., tachometer), which provides an electrical reference signal that contains

the fundamental frequency and all of the harmonics of the primary noise source. This technique has several

advantages compared to a broadband ANC system using acoustic reference sensors; (i) effect of feed-

back is eliminated (ii) aging and non-linearities associated with acoustic reference sensors are avoided,

(iii) causality condition is preserved due to periodicity. In narrowband ANC systems, once an electric

reference signal is available, a corresponding acoustic reference signal is internally generated to assist

the noise cancellation process.


8

x(n)

y(n) Σ

P

F

W

S

d(n)

e(n) x'(n)

Σ

M

M

M

J

J

K

Fig. 4. Broadband adaptive MIMO feed-forward ANC system with feed-back and secondary paths.

3) MIMO Feed-forward ANC: When the noise field of interest increases in size and bandwidth, the

number of active acoustic modes increases. To control multiple acoustic modes, it is necessary to use

multiple-channel ANC systems with multiple secondary sources, error signals and reference signals. A

MIMO feed-forward ANC system employs J reference sensors to observe the primary noise, M error

sensors to measure the residual noise and K secondary sources to produce the anti-noise. Figure 4

illustrates the block diagram of a broadband adaptive MIMO feed-forward ANC system with feed-back

and secondary path transfer functions. The wide arrows represent a flow of vectors (multi-channel acoustic

or electrical signals). The matrix P represents M×J primary path transfer functions, matrix S represents

K×M secondary path functions, matrix F represents K×J feed-back path functions and W represents

a matrix of K × J adaptive filters each serving an individual feed-forward channel.

B. Application of Feed-forward Control to Car Noise Cancellation

In the application of car-noise cancellation, feed-forward ANC systems are mostly applied for engine

noise cancellation. This is because it’s easier to obtain a reference signal directly from the engine resulting

in high coherence between the reference and error signals. Furthermore, as mentioned earlier, engine noise

is often periodic. Therefore, ANC for engine noise is often approached via a cost-effective narrowband

feed-forward system using an engine speed reference sensor, low-cost microphone error sensor(s) and

the vehicle’s in-built loudspeaker system as control sources. Such active noise control systems have been

commercially implemented by a number of manufactures as discussed later in Section VI.

Ideally, noise cancellation inside a vehicle would require the total acoustic energy distributed over

the entire global region to be minimized. Since this is an impractical task, the control region is often

sampled using one or more error sensors distributed over the control region. The total acoustic energy to


9

be minimized is then approximated by the sum of squares of the sensor output as

Jp =M∑m=1

p2m (3)

where pm is the sound pressure at the mth error sensor position (m = 1 · · ·M ). The effects due to the

above approximation is often comparable for low frequency control but gets increasingly noticeable at

high frequencies. The accuracy of the above approximation largely depends on the location of the error

sensors because as mentioned in section II-B, noise field characteristics inside an enclosure are largely

related to the enclosure’s active number of acoustic modes and structural-acoustic coupling. The effects of

structural-acoustic coupling on ANC inside vehicles have been thoroughly studied over the last 20 years

[1], [14], [15]. In [1] Elliot et.al showed that at low frequencies, a single sensor is capable of achieving

significant control, however with increasing complexity of the soundfield (frequency and geometry),

multiple acoustic modes become active and therefore, multiple sensors are required to successfully couple

into all of them. The frequency limit of global control was shown to be directly related to the modal

overlap or the number of acoustic modes that are significantly excited at a given frequency f , which

increases with cube of f . Therefore, in order to achieve control over the entire global region with a size

of a car, the amount of sensors required are often impractical.

In order to improve the control bandwidth, an alternative control strategy was to be developed. Such

a strategy was recently investigated in [6], [16], which attempts to control the soundfield within smaller

spatial regions (regional control), particularly surrounding the driver’s/passenger’s head. Regional or local

control of noise reduces the volume over which the noise energy has to be minimized and therefore

reduces the constraints on the control system and increases the control bandwidth. In [6], the authors

investigated a regional narrowband feed-forward system over two regions, one rectangular region across

the front seats and a second rectangular region across the rear seats. The performance of the system

was investigated through simulations and synthesis based on transfer functions measured in a rectangular

car cabin mock-up. The control system comprised of four secondary sources positioned at the standard

car audio loudspeaker positions, and eight error sensors positioned at the four head rest positions (two

sensors on each headrest). The acoustic potential energy within the control regions were shown to be

significantly reduced at frequencies up to 370 Hz. This is around twice the control bandwidth of global

feed-forward control using a similar system. The authors mention a potential issue with the regional

feed-forward control strategy, that is the system is said to be susceptible to unobservable modes that

result in enhancements in the regional acoustic potential energy. However, it has also been shown that

these effects can be limited by using control effort weighting parameters.


10

Feedback ANC

y(n) e(n)

Primary Noise

Error Microphone Loudspeaker

Noise Source

Fig. 5. A SISO adaptive feed-back ANC system.

In addition to engine noise cancellation, feed-forward systems are also applied in road noise cancella-

tion. In [17], Oh et.al presented a leaky constraint MIMO feed-forward ANC system for road booming

noise control in a mid-size passenger vehicle using 2 accelerometers, 2 control loudspeakers and a

single error microphone. Based on experimental results, the optimum positioning for the above devices

were chosen, and during driving tests on rough asphalt and turtle/back roads at 60 km/h, a reduction

of 6−dB A−weighted sound pressure level (A−weighting accounts for the relative loudness perceived

by the human ear) in the road booming noise was achieved. Due to the high cost of accelerometers,

feed-forward ANC is generally regarded as unsuitable for mass production, however with the recent

introduction of low-cost MEMS accelerometers, it is expected to change.

V. FEED-BACK CONTROL SYSTEMS

This section discusses the adaptive feed-back control systems used for broadband ANC. Unlike feed-

forward systems, feed-back systems directly employ the signal(s) from error sensor(s) to drive the

secondary source(s) via a controller. Since the error sensor signal is fed back to the secondary source, the

system cannot be optimized on a frequency-by-frequency basis as in feed-forward control, and therefore,

the whole frequency response (broadband) must be considered at all times. The performance of feed-back

control systems are limited by their stability, which is largely dependent on the system delay. Therefore,

the control bandwidth of feed-back control systems are inversely proportional to the spacing in between

the error sensor(s) and secondary source(s).

A. Feed-back Control Algorithms

1) SISO Feed-back Control: A single channel adaptive feed-back ANC system, as shown in Fig 5,

was first proposed in [18]. Based on the Internal Model Control (IMC) architecture, an adaptive feed-

back system can be viewed as an adaptive feed-forward system, that synthesizes or regenerates its own

reference signal using the error signal and the adaptive filter output. The basic concept of this model is


11

y(n)

d(n)

LMS

Σ

y'(n)

e(n)

Ŝ(z)

x'(n)

W(z)

Σ

S(z)

Ŝ(z)

x(n)

d'(n)

Fig. 6. A SISO adaptive feed-back ANC system using the IMC architecture.

to estimate the primary noise d(n) present at the error sensor and use it as a reference signal x(n) for

the adaptive filter. If the secondary path transfer function S(z) is known, the primary noise signal d(n)

can be synthesized using

X(z) ≡ D(z) = E(z) + S(z)Y (z) (4)

where the notation ˆ represents an estimated value. Therefore, the reference signal synthesis technique

filters the secondary source signal y(n) using the secondary path estimate S(z) and combines it with

e(n) to regenerate the primary noise. Figure 6 shows a complete SISO feed-back ANC system using

the FXLMS algorithm with secondary path cancellation as discussed in section IV-A1. When applying a

feed-back control system, the overall control system stability is very important next to the noise reduction

level. This is generally analyzed by checking the Nyquist stability criterion, which states that the polar

plot of the open-loop response must not enclose the Nyquist point (−1, 0) as ω increases from −∞ to

+∞ [19]. In a practical system, since the open-loop response often varies with time, it is typical to set

the feed-back gain to stabilize the system despite its variations. A system that stabilizes with such a gain

is said to have robust stability.

2) MIMO Feed-back Control: The SISO feed-back ANC system is extendable to a multiple-channel

system with K secondary sources and M error sensors. Such a system will have M×K secondary paths.

Each path Smk(z) is from the kth secondary source to the mth error sensor, and needs to be estimated

by a filter Smk(z). These estimated filters along with the K secondary/control signals yk(n) and the M

error signals em(k) will synthesize M reference signals xm(n) for the corresponding K ×M adaptive

filters Wkm(z). A multiple channel FXLMS algorithms will be required to calculate the coefficients of

the adaptive filters. Figure 7 illustrates a block diagram of the entire process described above. In practice,

the extension of a SISO feedback system to a MIMO feedback system is, somewhat complex due to the

need to calculate the eigenvalues of the open-loop response in order to assess the controller stability.


12

K x M Adaptive

Filters

Reference Signal

Synthesizer

x(n)

y(n)

e(n)

M

K

FXLMS

Noise Source

y(n)

M

e1(n)

eM(n)

y1(n)

yk(n)

Fig. 7. A MIMO adaptive feed-back ANC system using M error sensors and K secondary sources.

B. Application of Feed-back Control to Car Noise Cancellation

Feed-back control is predominantly used to minimize the effects of road noise. Sano et al [5] designed

and implemented a SISO feed-back control system for boom noise control of a Honda station wagon.

The system mainly attempts to control the boom noise at 40 Hz present in the front seats of the car,

which is due to the first acoustic longitudinal mode of the vehicle’s enclosure. The feed-back system

employs a single error microphone, positioned under the front seat and the two front door loudspeakers

of the car’s in-built speaker system (the two speakers are driven in-phase, hence the control system is

SISO) to achieve noise reduction up to 10 dB. The authors observed an undesired side-effect in the

rear sears, where the boom noise was increased by 3 dB. To avoid this, they proposed a simultaneous

fixed feed-forward control system, which uses the previous system’s error microphone as a reference

signal to minimize the boom noise present at the rear seats. The secondary system utilized the two rear

door loudspeakers from the vehicle’s in-built speaker system driven in phase. This approach managed

to achieve a 10 dB reduction of the boom noise at the front seats while avoiding the increase of sound

level at the rear seats, where the boom noise is not significant.

Similar to feed-forward control, the performance of SISO feed-back control is largely limited by

increasing frequency and the enclosure size. The theory involved with employing MIMO feed-back

control for road noise control inside automobiles was first presented by Elliot and Sutton in [4]. Recent

work on this approach including a practical investigation was carried out by Cheer et al in [6], [20],

where the authors utilized a non-rigid car cabin mock-up. With 8 error sensors and 4 control sources,

the MIMO system required a total of 32 FIR filters. The authors simulated road noise using uncorrelated

structural vibrations for which the system managed to cancel an increased number of acoustic modes


13

compared to the modal feed-back controller. When the road noise was simulated using uncorrelated point

sources, the system was capable of canceling all the active modes up to a control bandwidth of 100

Hz. The performance of the above system in a practical automobile environment was also investigated

by Cheer et al [6], [19] inside a small city car. In this work, the authors utilized 16 error microphones

(8 on the floor, with a pair near each tire and 8 on the seats, with a pair on each head-rest) and the

4 in-built door loudspeakers as control sources. The system performance was synthesized offline based

on the transfer function measurements inside the car. The authors managed to achieve significant noise

reduction up to 8 dB in the low frequency range and an average reduction of 3 dB between 80− 200 Hz

where the road noise is prominent.

As discussed in section IV-B, a recent approach [21] to improve control bandwidth and reduce system

complexity is to simplify the global control requirement by regional control where the control region is

shrunk to a small area around the head position(s). In [22], the authors implemented a regional feed-back

control system in a Ford S-Max employing a horizontal grid array of 25 error microphones positioned in

front of the headrest on the front passenger seat and two control loudspeakers mounted on the headrest.

For 80 kmph smooth driving conditions, the regional control system achieved noise reduction up to 300

Hz. Also when the error was only averaged over 4 microphones close to typical ear positions, the control

bandwidth was extended up to 500 Hz revealing the potential advantages of regional control systems.

VI. COMMERCIAL SYSTEMS

In the commercial automotive space, noise control is still predominantly achieved via passive control.

However, to overcome limitations related to passive control, more companies are increasingly applying

active control to mass production. Commercial active control systems are typically applied for both

noise and vibration control. While ANC utilizes an acoustic system, AVC typically comprises of an

active engine mount (ACM), which not only reduces vibrations but also reduces noise inside the cabin.

More recently, the concept of active sound control (ASC) was introduced to commercial automotive

solutions, which improves the driving experience by synthesizing certain sounds that are essential for

the perceptual sound quality inside/outside the car. This process, which is similar to ANC, typically

uses adaptive algorithms to change the coefficients of a set of digital filters such that not only are some

selected frequencies canceled by secondary loudspeaker(s) generating an inverse disturbance signal(s),

but others are controlled to a predetermined level, or even enhanced [12], [23]. In this section, we present

the chronological advancement of active control in commercial automotive applications with the main

focus on ANC.


14

Research and development of ANC became popular in the latter half of 1980 [1]. The earliest ANC

systems were feed-forward arrangements based on MIMO FxLMS for tonal engine noise (or booming

noise) control inside cars [24]. Implementation of such a system was initially carried out in collaboration

with University of Southampton and Lotus Engineering, where 4 loudspeakers were adjusted at the

engine’s firing frequency and its harmonics to minimize the mean-square pressure at 8 error microphones

located on the head-rests [25], [26]. ANC in mass production vehicles was first introduced by Nissan in

1991 [27] for booming noise, where a limited grade Nissan mid-size car was optionally installed with

a separate ANC system, that consisted of additional loudspeakers, microphones and a MIMO FxLMS

control system. The cost of implementation was quite high and the resulting level of noise reduction was

not deemed to be significant. Therefore, at the time, it was not generally accepted as a useful technology.

In addition to engine booming noise reduction, research has also been actively carried out on road

noise reduction since early 1990s [28], [29]. In 2000 Honda introduced ANC for low-frequency narrow

band road noise control. It was was applied as standard equipment in a station wagon, where a fixed

feedback controller based on control engineering theory was utilized [5].

Nearly a decade after Nissan’s attempt, commercial interest in engine booming noise reduction started

re-gaining attention due to the integration of the ANC system to the vehicle’s in-built audio system.

In 2003, Honda introduced ANC for booming noise caused by the Honda V6 engine model, which

employed the variable cylinder management (VCM) technology to improve fuel economy by providing

three-cylinder operation [30], [31]. This ANC system employed an adaptive notch filter based MIMO feed-

forward controller and was combined with an active control engine mount (ACM) to reduce vibrations.

Currently all VCM engine models from Honda are equipped with ANC and ACM. In 2006, Honda

combined an ASC system with their existing ANC solution for engine booming noise control. The ASC

system was introduced to improve the internal cabin sound by synthesizing engine acceleration sounds

for speeds above 2500 rpm such that it delivers a sporty feel to the driver [32]. In 2008, both Toyota

[33] and GM introduced ANC for booming noise in a mid-size car and a mid-size SUV respectively.

Both solutions were based on adaptive MIMO feed-forward controllers. Soon after in 2009, Nissan re-

introduced a MIMO ANC system based on adaptive feed-forward control for engine booming noise in

a mid-size car [34]. In 2011, Honda introduced commercial solutions for low frequency road noise by

integrating an extra feedback controller (adaptive notch filter) to their existing booming noise controller

(MIMO feed-forward) [35]. In the frequency range below 100 Hz, this system is claimed to achieve

10 dB reduction of noise level inside a mid-size car [36] (See Fig. 8). The aforementioned road noise

controller by Honda was recently (2015) updated with an expanded low frequency range [37].

In addition to the ANC solutions provided by automobile manufacturers, leading audio system de-


15

Front seats Rear seats

600 1200 1800 2400 600 1200 1800 2400 (rpm) (rpm)

Acou

stic

Pr

essu

re (d

B)

Fig. 8. Noise reduction levels of the HONDA ANC system at front seats and rear seats, adopted from [35], [36].

velopers such as Bose and Harman have also developed noise management solutions for automobiles.

The Bose R© Active Sound Management System (ASM) is an example for such a solution [38]. The

main technologies used in ASM are Bose Engine Harmonic Cancellation (EHC), Bose Engine Harmonic

Enhancement (EHE) and Bose Rapid Mode Transition (RMT). The EHC technology is an ANC solution

that minimizes booming noise utilizing a feed-forward control system [39]. The EHE and RMT systems

are both ASC solutions that synthesize artificial sound to improve the driving experience. The EHE

technology provides desirable linear sounds (or sporty) by masking sound anomalies that occur during

acceleration [40] while the RMT technology provides additional control parameters to synthesize a seam-

less sound experience during variable powertrain and cabin modes (e.g., cylinder deactivation/reactivation,

hybrid operation) [38]. Bose ASM was previously available only for vehicles with Bose sound system

hardware however, since 2013, it was released as a software solution integrated in a chip for global auto

manufacturers. Currently, Bose ASM is integrated in vehicles manufactured by GM, Nissan, Audi [41],

Porsche etc., particularly in their luxury divisions.

Introduced recently in 2015, HALOsonicTM is a another commercially available noise management

solution provided by Harman International and Lotus Cars, which comprises of a suite of four tech-

nologies to enhance the in-car audible environment and improve pedestrian safety [42], [43]. The two

technologies directly related to ANC are the Road Noise Cancellation (RNC) system and the Engine

Order Cancellation (EOC) system. The RNC system is a broadband feed-forward control system with

accelerometers as reference sensors. It is based on road noise cancellation solution originally presented

in [28]. The HALOsonicTM EOC system, is a combined system with feed-forward control to reduce noise

due to engine rotations and a feed-back controller to reduce noise due to internal combustion engine and

exhaust components. Note that the design and specifications of the above systems may vary based on the

size, shape and cost of the vehicle model of interest. The remaining two technologies of HALOsonicTM

focus on ASC or electric sound synthesis in quiet cars (e.g., electric cars, hybrid cars). This is done in


16

two areas; (i) internally, to improve the passenger experience and (ii) externally, to improve the safety

of pedestrians. The internal (iESS) system helps synthesize an exhilarating engine sound that adds a

very emotional element to the overall driving experience. It helps reinstate original engine sound in case

of sound loss due to OEM features such as downsized engines and the use of turbochargers. iESS also

offers multiple engine sound modes (for example normal, moderate and sporty engine sound) for the same

car, thereby enhancing car occupants’ emotional experience. The external (eESS) system mainly provides

safety to quieter cars, where active system is optimized to operate in urban environments with the greatest

risk of a collision with pedestrians, especially high-risk groups such as the elderly, children, cyclists, and

particularly the blind and their guide dogs. eESS helps automakers comply with governmental safety

regulations. The sound of a car engine is an integral part of the experience behind the wheel and plays a

crucial role in defining the DNA of the car. eESS is capable of creating custom-designed engine sounds,

thereby helping to retain an OEM-specific (Original Equipment Manufacturer) sound DNA for the car

[42].

VII. PRACTICAL LIMITATIONS OF CURRENT ANC SYSTEMS

In this section, we discuss the main limitations related to mass production of ANC systems inside auto-

mobiles. As reviewed in [44], these include (i) effects due to constrained number of microphones/speakers,

(ii) stability issues in feed-back control systems (iii) system latency (iv) uncertainties (v) ANC system

integration issues and (vi) system production tuning issues. A brief discussion on each of the above

constraints are as follows. Due to cost restrictions, the numbers of speakers and microphones employed

in ANC systems are limited. Currently, a typical production set-up consists of 4 microphones and 4− 5

speakers, which can only achieve global noise control over 30− 250 Hz (the lower limit is determined

by the speaker characteristics whereas the higher limit is determined by vehicle interior and component

placement). Control up to at least 200 Hz is often desired because 200 Hz is equivalent to the firing

frequency of an I4 engine at 6000 RPM, which is the dominant cause of booming noise. The active

number of engine orders of a typical automobile is however much higher, and with new technologies like

cylinder deactivation, there exist extra noise components that need to be addressed. Therefore, improved

ANC performance requires increased numbers of microphones/speakers with increased computational

requirements, memory, and higher tuning efforts due to the increased complexity.

Another practical constraint that effects the the performance of ANC systems, particularly employing

feedback control loops, is stability. To minimize side-effects due to stability, it is important to carefully

calculate the sensitivity function of the closed loop. At present, there exist advance modeling techniques

to understand the system behavior through computer aided technology [41], which is expected to make


17

significant progress in the coming years. Within the working frequency range of an ANC System, another

practical limitation that arises is system latency. This is the signal latency added by processes like A/D

and D/A conversion, and DSP processing latency. It is a difficult task to pin-point the exact instance when

system latency starts to deteriorate the ANC performance. In practice, an ANC system latency (latency in

the microphone input-ANC processing-speaker output loop) of 2 ms is considered to be acceptable while

3 ms is the upper limit to avoid significant degradation [44]. ANC performance in realistic automobile

cabins, is often affected by uncertainties such as number and placement of seated passengers, opening

and closing of windows/doors, and vehicle interior production tolerances. To ensure consistent ANC

performance, it is important to improve system robustness while minimizing uncertainties. The typical

approach to achieve this is via measuring and modeling different components of overall uncertainty as

accurately as possible under realistic conditions, and setting the ANC system parameters to guarantee

robustness.

In addition to the aforementioned concerns, another key aspect that affects the implementation of ANC

in production vehicles is integration of the ANC system to the existing audio system for parallel usage.

Initially, ANC solutions were added as an extra control unit (DSP audio amplifier) causing no impact

on the existing audio system, however it was soon deemed to be ineffective with regard to cost, weight

and space. A subsequent ANC solution was to add some dedicated processing resources for ANC into

the existing audio system (i.e, plug-in module for head unit with dedicated DSP), which omits the need

for an additional control unit while minimizing added weight and space. A more recent and improved

ANC solution is to fully integrate ANC in the form of software into the existing audio system. This

is done by (i) adding ANC as a software on the amplifier without extra processing unit (e.g., Analog

devices’ SHARC processor) or (ii) integrating functional software using System on Chip (SOC) solutions

(e.g., NXP Chip for the Bose Active Sound Management System). While the commercial application of

the above solutions are still limited, they are expected to be utilized more broadly in the near future.

One more issue that affects ANC implementation in mass production vehicles is system tuning during

production. This involves the measurement of secondary path transfer functions and determination of

algorithm parameters such as the number of engine orders to cancel. With uncertainty issues mentioned

above, and multiple available powertrains, it is important to tune the ANC system for each of the vehicle

variants. This task requires a lot of time and man power and with increasing demand to shorten the

vehicle development period, there is an urgent need to opt for advanced CAE technologies that enable

faster tuning.


18

VIII. SPATIAL SOUNDFIELD CONTROL IN ACTIVE NOISE CONTROL

Up to this point, we have only discussed ANC techniques that model the noise field in terms of acoustic

modes and structural-acoustic coupling. By now, it is common knowledge that the aforementioned ANC

is effective at low frequencies, but have limitations at high frequencies due to increased requirement of

microphones/speakers and related cost. Recently, research has been carried out to model noise fields in

an alternative domain such that characteristics like sparsity can be exploited to bring down the minimum

requirement of microphones/speakers. This concept is based on spatial soundfield control and initial

research on this topic is described below.

Spatial soundfield control involves acoustic control over a continuous spatial region utilizing a finite

set of transducers distributed over the region of interest. Two well developed techniques to achieve spatial

soundfield control are wavefield synthesis [45] and higher order ambisonics (HOA) [46]. Currently, HOA

is the only technique utilized for spatial noise cancellation inside automobiles, and therefore, the overview

given in this section will be limited to HOA. HOA is conceptually based on the cylindrical/spherical

harmonics based solution to the wave equation. This solution represents the incident pressure at any

arbitrary point x within a control region of radius R, with respect to its origin by [47]

p(x, k) =

N∑

n=−Nαn(k)Jn(kr)e

inφ 2D region with x = (r, φ)

N∑n=0

n∑m=−n

αnm(k)jn(kr)Ynm(θ, φ) 3D region with x = (r, θ, φ)

(5)

where k = 2πf/c represents the wave number with f and c representing frequency and speed of sound

respectively, α denotes the HOA harmonic coefficients, Jn(·) and jn(·) represent the cylindrical and

spherical Bessel functions of order n, respectively, Ynm(·) denotes the spherical harmonic function and

N = dkRe is the summation’s truncation limit (commonly referred to as soundfield order) derived based

on inherent properties of Bessel functions. The main advantage of the above decomposition is it gives

the ability to record or produce an entire continuous spatial soundfield by considering only a finite set

of coefficients. When recording a spatial soundfield, these coefficients have a direct relationship with

the microphone outputs in the form P = Tα where α is a vector of recorded soundield coefficients,

T is a transformation matrix and P is a vector of microphone recordings. Similarly, when producing a

soundfield, the above coefficients have a direct relationship with the loudspeaker driving signals in the

form α = T 1W , where α is now a vector of desired soundield coefficients, T 1 is a transformation

matrix and W is a vector of loudspeaker driving signals. Generally, when recording/producing an N th

order soundfield, there exists a minimum requirement of (2N +1) or (N +1)2 sensors/loudspeakers for

2D and 3D soundfields, respectively. This is to avoid the undesired effects of spatial aliasing.


19

Fig. 9. Hardware setup for regional and spatial feed-back control.

When HOA based spatial soundfield control is occupied in active noise control, the residual field,

the noise field and the secondary soundfield are first decomposed in to cylindrical/spherical harmonic

coefficients. For example, in 3D ANC, the frequency transform of (6) is decomposed into

enm(k) = d(k)nm + y′(k)nm. (6)

The input and output of the adaptive controller are therefore spherical harmonic coefficients rather than

the direct error sensor outputs or control speaker weights. As a result, the standard block diagram for

feed-forward and feed-back systems needs to be updated by additional modal transformation blocks. In

[48], Spors et al designed and simulated a 2D HOA based massive feed-forward ANC system with 80

reference sensors, 80 error sensors and 80 loudspeakers, mainly for use in room noise cancellation. A

2D feed-back control system following the HOA technique was recently proposed by Zhang et al using

11 error sensors and 11 control sources [49]. This feed-back system was later extended with a sparse

FXLMS controller, particularly for the use in spatially sparse noise fields [50]. The main advantage

of HOA based ANC as observed in both feed-back and feed-forward systems mentioned above is the

significant improvement of convergence time and the significant decrease of spatially averaged residual

signal energy. However, due to the relationship N = dkRe, the minimum requirement of sensors/speakers

to control a sizable enclosure is impractically high, especially if the system is to be utilized for noise

reduction inside automobiles.

A. Application of Spatial Soundfield Control to Car Noise Cancellation

In [16], Chen et al investigated the applicability of spatial regional control in ANC inside automobiles.

The main purpose of the study was to derive the performance bounds of a feed-back system with the

automobile’s in-built speakers utilized as control sources. Results were synthesized using a fixed offline


20

Frequency (Hz)

50 100 150 200 250 300 350 400 450 500

No

ise

Po

we

r A

tte

nu

atio

n (

dB

)

-40

-35

-30

-25

-20

-15

-10

-5

Busy Road

Engine Only

AC Only

Freeway

Fig. 10. Performance bounds of spatial ANC : Noise power spectrum attenuation for 4 different driving conditions.

system based on transfer functions and noise measurements done at the front-left headrest of a Ford

Falcon XR6 (see fig. 9). The sensor array used was a commercially available 32-microphone spherical

array (Eigenmike). While the previously discussed spatial ANC systems preferred a spherical array of

control sources, the proposed system simplified this constraint to the vehicle’s own audio system, based

on a novel model for the primary noise field. This model was derived utilizing the spherical harmonic

decomposition of the recorded noise field such that it represents the primary noise field in terms of an

alternative set of basis functions. Based on a diverse set of noise measurements obtained inside the car

(e.g., engine only, AC only, road noise at specific speeds) the authors, found out that the noise field

inside a vehicle is generally sparse in terms of the proposed noise model. In fact, for the head-sized

region of interest, it was observed that only a single noise mode was active at all times. Therefore, it was

predicted that the vehicle’s in-built 2 channel audio system (Ford Falcon XR6 has 4 loudspeakers with

stereo control) will be sufficient to attenuate the active noise mode. Figure 10 shows the noise reduction

observed over frequency for 4 different driving conditions, where it’s observed that noise reduction is

relatively consistent with the attenuation levels varying between 35− 15 dB across 50− 500 Hz. These

results are quite promising in terms of the potential use of a vehicle’s own audio system for effective

ANC. A robustness analysis of this system however is still to be carried out. The theory of the above

design was later extended to support multiple-region ANC control and tested in the same vehicle [51].

From this investigation the authors concluded that a vehicle’s integrated loudspeakers when used as a

stereo system, are only capable of canceling the noise field up to 200 Hz at the head positions of two

seats simultaneously. To achieve similar reductions over 4 headrest positions simultaneously, a minimum

of four individually driven loudspeakers were said to be needed.


21

IX. SUMMARY AND FUTURE DIRECTIONS

In this article, a compact tutorial of ANC techniques was presented with a review of their application in

reducing undesired noise inside automobiles. Some of the recent advances have demonstrated significant

improvements in the noise reduction levels as well as the cost and implementation complexity. While

the techniques discussed above may individually focus on a particular noise field (e.g., road noise only,

engine noise only), it is proven through research and commercial products that a combination of these

strategies can deliver significant benefits in realistic conditions.

Future opportunities for improving in-car ANC exist in (i) cost reduction (ii) practical implementation

and commercialization of regional sound field control (iii) integration of regional ANC with future in-car

infotainment systems and (iv) researching on alternative noise modeling techniques to bring down the

system components. At present, main drivers for cost are hardware components, particularly the extra

requirement for error/reference microphones and control loudspeakers. With the introduction of MEMS

microphones, and MEMS loudspeakers (e.g., Audio Pixels [52]), technically feasible low-cost ANC sys-

tems could be introduced, possibly with better performance. As mentioned earlier, regional ANC is a well

researched topic that could reduce the overall system requirements. Regional ANC can be also extended

for multiple-regions serving individual passengers. Practical implementation and commercialization of

these solutions are still minimal and has potential to reduce costs and improve efficiency. With the

current global trend of instant connectivity, vehicles are evolving to provide infotainment systems rather

than just radio. These include the availability of different wireless interfaces including WiFi, Bluetooth

which forces all of the systems to move to digital. Future In-vehicle infotainment systems are predicted

to be comprised of "center stack computers” [44] to process all types of media content and a network

hub to serve multiple media/data streams, possibly on a per/seat, per/display basis. With the introduction

of such systems, its essential for ANC to be re-introduced with appropriate low-latency audio processing

that handles digital signals. Finally, another important future direction that could improve ANC efficiency

is by looking for alternative modeling methods for the noise field inside the car. Even though the current

ANC systems are largely restricted to low frequencies, an alternate model that describes noise fields in

terms of a lower number of active modes may significantly enhance the system performance for the same

number of microphones/speakers, specially when the noise field is directionally sparse.

REFERENCES

[1] P. A. Nelson and S. J. Elliott, Active control of sound. Academic press, 1991.

[2] G. Cerrato, “Automotive sound quality - powertrain, road and wind noise,” Sound and Vibration, vol. 43, pp. 16–24, April

2009.


22

[3] S. M. Kuo and D. R. Morgan, “Active noise control: a tutorial review,” Proceedings of the IEEE, vol. 87, no. 6, pp.

943–973, 1999.

[4] S. J. Elliott and T. J. Sutton, “Performance of feedforward and feedback systems for active control,” IEEE Transactions

on Speech and Audio Processing, vol. 4, no. 3, pp. 214–223, 1996.

[5] H. Sano, T. Inoue, A. Takahashi, K. Terai, and Y. Nakamura, “Active control system for low-frequency road noise combined

with an audio system,” IEEE Transactions on Speech and Audio Processing, vol. 9, no. 7, pp. 755–763, 2001.

[6] J. Cheer, “Active control of the acoustic environment in an automobile cabin,” Ph.D. dissertation, University of Southampton,

UK, 2012.

[7] D. Thompson and J. Dixon, Advanced Applications in Acoustics, Noise and Vibration. London: Spon Press, 2004, ch.

Vehicle noise, p. ch. 6.

[8] D. Robinette, M. Grimmer, J. Horgan, J. Kennell, and R. Vykydal, “Torque converter clutch optimization: improving fuel

economy and reducing noise and vibration,” SAE International Journal of Engines, vol. 4, no. 2011-01-0146, pp. 94–105,

2011.

[9] M. R. Schroeder, “Frequency-correlation functions of frequency responses in rooms,” The Journal of the Acoustical Society

of America, vol. 34, no. 12, pp. 1819–1823, 1962.

[10] S. J. E. andM. E. Johnson, “Radiation modes and the active control of sound power,” Journal of the Acoustical Society of

America, vol. 94, pp. 2194–2204, 1993.

[11] S. M. Kim, “Active control of sound in strucutral-acoustic coupled systems,” Ph.D. dissertation, University of Southampton,

UK, April 1998.

[12] H. Sano, “Modern advancements in passive and active noise and vibration control technology in automobiles,” in Proc.

InterNoise, 2011.

[13] D. R. Morgan, “An analysis of multiple correlation cancellation loops with a filter in the auxiliary path,” IEEE Transactions

on Acoustics, Speech and Signal Processing, vol. 28, no. 4, pp. 454–467, 1980.

[14] J. Pan, C. Hansen, and D. Bies, “Active control of noise transmission through a panel into a cavity: I. analytical study,”

The Journal of the Acoustical Society of America, vol. 87, no. 5, pp. 2098–2108, 1990.

[15] J. I. Mohammad, “The active control of random sound inside cars,” Ph.D. dissertation, University of Southampton, 2006.

[16] H. Chen, P. Samarasinghe, T. D. Abhayapala, and W. Zhang, “Spatial noise cancellation inside cars: Performance analysis

and experimental results,” in 2015 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA).

IEEE, 2015, pp. 1–5.

[17] S.-H. Oh, H.-s. Kim, and Y. Park, “Active control of road booming noise in automotive interiors,” The Journal of the

Acoustical Society of America, vol. 111, no. 1, pp. 180–188, 2002.

[18] L. Eriksson, “Recursive algorithms for active noise control,” in Proc. Int. Symp. Active Control of Sound Vibration, 1991,

pp. 137–146.

[19] J. Cheer and S. J. Elliott, “Multichannel control systems for the attenuation of interior road noise in vehicles,” Mechanical

Systems and Signal Processing, vol. 60, pp. 753–769, 2015.

[20] J. Cheer and S. Elliott, “Mutlichannel feedback control of interior road noise,” in Proceedings of Meetings on Acoustics,

vol. 19, no. 1. Acoustical Society of America, 2013, p. 030118.

[21] S. J. Elliott, W. Jung, and J. Cheer, “The spatial properties and local active control of road noise,” in Proc. of Euro-noise,

2015.

[22] J. Cheer, S. Elliott, and W. Jung, “Sound field control in the automotive environment,” 3rd International ATZ Automotive

Acoustics Conference, 2015.


23

[23] L. E. Rees and S. J. Elliott, “Adaptive algorithms for active sound-profiling,” IEEE Transactions on Audio, Speech, and

Language Processing, vol. 14, no. 2, pp. 711–719, 2006.

[24] S. Elliott, I. Stothers, P. Nelson, A. McDonald, D. Quinn, and T. Saunders, “The active control of engine noise inside

cars,” in INTER-NOISE and NOISE-CON Congress and Conference Proceedings, vol. 1988, no. 3. Institute of Noise

Control Engineering, 1988, pp. 987–990.

[25] S. J. Elliott, “A review of active noise and vibration control in road vehicles,” ISVR Technical Memorandum, 981, 2008.

[26] A. McDonald, S. Elliott, and M. Stokes, “Active noise and vibration control within the automobile,” in Proc. Int. Symp.

Active Control of Sound Vibration, 1991, pp. 147–156.

[27] S. Hasegawa, T. Tabata, A. Kinoshita, and H. Hyodo, “The development of an active noise control system for automobiles,”

SAE Technical Paper, Tech. Rep., 1992.

[28] T. J. Sutton, S. J. Elliott, A. M. McDonald, and T. J. Saunders, “Active control of road noise inside vehicles,” Noise Control

Engineering Journal, vol. 42, no. 4, 1994.

[29] R. J. Bernhard, “Active control of road noise inside automobiles,” in INTER-NOISE and NOISE-CON Congress and

Conference Proceedings, vol. 1995, no. 5. Institute of Noise Control Engineering, 1995, pp. 21–32.

[30] T. Inoue, A. Takahashi, H. Sano, M. Onishi, and Y. Nakamura, “NV countermeasure technology for a cylinder-on-demand

engine-development of active booming noise control system applying adaptive notch filter,” SAE Technical Paper, Tech.

Rep., 2004.

[31] H. Matsuoka, T. Mikasa, and H. Nemoto, “NV countermeasure technology for a cylinder-on-demand engine–development

of active control engine mount,” SAE Technical Paper, Tech. Rep., 2004.

[32] Y. Kobayashi, T. Inoue, , A. Takahashi, and K. Sakamoto, “Active sound control in automobiles,” in INTER-NOISE and

NOISE-CON Congress and Conference Proceedings, vol. 2008, no. 4. Institute of Noise Control Engineering, 2008, pp.

5001–5009.

[33] H. Morita, T. Mochizuki, and H. Yoshida, “Development of active noise control-mechanism analysis and countermeasure

of the waterbed phenomena,” in Society of Automotive Engineers of Japan Annual Congress, no. 20095561. Society of

Automotive Engineers of Japan, 2009, pp. 15–18.

[34] N. Masahiro, O. Daisuke, H. N. Yoshiharu Nakaji, M. Masahiro, and H. Mineura, “Development of anc technology for

reconciling acceleration sound with fuel economy and quietness,” in Society of Automotive Engineers of Japan Annual

Congress, no. 20105357. Society of Automotive Engineers of Japan, 2010, pp. 7–10.

[35] A. Takahashi, T. Inoue, K. Sakamoto, and Y. Kobayashi, “Integrated active noise control system for low-frequency noise

in automobiles,” in INTER-NOISE and NOISE-CON Congress and Conference Proceedings, vol. 2011, no. 6. Institute

of Noise Control Engineering, 2011, pp. 2105–2113.

[36] HONDA New Zealand. (2014) Honda active noise cancellation. [Online]. Available:

http://www.honda.co.nz/technology/driving/anc/

[37] K. Sakamoto and T. Inoue, “Development of feedback-based active road noise control technology for noise in multiple

narrow-frequency bands and integration with booming noise active noise control system,” SAE International Journal of

Passenger Cars-Mechanical Systems, vol. 8, no. 2015-01-0660, pp. 1–7, 2015.

[38] Bose Corporation. (2016) Bose active sound management technology. [Online]. Available:

http://www.bose.com/prc.jsp?url=/automotive/innovations/asm.jsp

[39] A. GANESHKUMAR, “Engine harmonic cancelling system and operating method thereof,” September 2011. [Online].

Available: http://www.sumobrain.com/patents/WO2011112417A1.html

[40] C. M. Hera, “Engine harmonic enhancement control,” Nov. 3 2015, uS Patent 9,177,544.


24

[41] R. Schirmacher, R. Kunkel, and M. Burghardt, “Active noise control for the 4.0 tfsi with cylinder on demand technology

in audi’s s-series,” SAE Technical Paper, Tech. Rep., 2012.

[42] HARMAN International. (2016) Halosonic noise management solutions. [Online]. Available: http://www.halosonic.co.uk/

[43] ——. (2015) Halosonic: Fututre of sound. [Online]. Available: http://harmaninnovation.com/blog/halosonic-future-sound/

[44] R. Schirmacher, “Current status and future developments of anc systems,” SAE International Journal of Passenger Cars-

Mechanical Systems, vol. 8, no. 2015-01-2223, pp. 892–896, 2015.

[45] S. Spors, R. Rabenstein, and J. Ahrens, “The theory of wave field synthesis revisited,” in 124th AES Convention, 2008,

pp. 17–20.

[46] D. B. Ward and T. D. Abhayapala, “Reproduction of a plane-wave sound field using an array of loudspeakers,” IEEE

Transactions on Speech and Audio Processing, vol. 9, no. 6, pp. 697–707, 2001.

[47] E. Williams, Fourier Acoustics: Sound Radiation and Nearfield Acoustic Holography. London, UK: Academic Press,

1999, pp. 115–125.

[48] S. Spors and H. Buchner, “Efficient massive multichannel active noise control using wave-domain adaptive filtering,” in

3rd International Symposium on Communications, Control and Signal Processing, 2008. ISCCSP 2008. IEEE, 2008, pp.

1480–1485.

[49] J. Zhang, W. Zhang, and T. D. Abhayapala, “Noise cancellation over spatial regions using adaptive wave domain processing,”

in 2015 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA). IEEE, 2015, pp. 1–5.

[50] J. Zhang, T. D. Abhayapala, P. Samarasinghe, W. Zhang, and S. Jiang, “Sparse complex fxlms for active noise cancellation

over spatial regions,” in International Conference on Acoustics, Speech and Language Processing (ICASSP). IEEE, 2016.

[51] H. Chen, P. Samarasinghe, and T. D. Abhayapala, “In-car noise field analysis and multi-zone noise cancellation quality

estimation,” in 2015 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA).

IEEE, 2015, pp. 773–778.

[52] Audio Pixels Limited. (2016) Audiopixels technology. [Online]. Available:

http://www.audiopixels.com.au/index.cfm/technology/

Prasanga Samarasinghe ([email protected]) received the B.E. degree (with first class

honors) in Electronic and Electrical Engineering from the University of Peradeniya, Sri Lanka in 2009.

She completed the Ph.D. degree at the Australian National University (ANU), Canberra in 2014. She is

currently working as a Research Fellow at the Research School of Engineering at ANU, and her research

interests include spatial audio, active noise control and multi-channel signal processing.


25

Wen Zhang ([email protected]) received the B.E. degree in telecommunication engineering from

Xidian University, Xi’an, China, in 2003, the M.E. degree (with first class honours) and the Ph.D. degree

from the Australian National University, Canberra, in 2005 and 2010, respectively. From 2010 to 2012,

she was a Postdoctoral Fellow at CSIRO Process Science and Engineering in Sydney, Australia. She

is currently working as a Research Fellow in the College of Engineering and Computer Science at the

Australian National University. Her research interests include spatial audio, binaural source localization,

and active noise control. She is awarded the Discovery Early Career Researcher Award (DECRA) fellowship by the Australian

Research Council (ARC) in 2015.


26

Professor Thushara Abhayapala ([email protected]) received the B.E. degree (with Hon-

ors) in Engineering in 1994 and the Ph.D. degree in Telecommunications Engineering in 1999, both from

the Australian National University (ANU), Canberra. Currently he is the Deputy Dean of the College of

Engineering & Computer Science, ANU. He was the the Director of the Research School of Engineering

at ANU from January 2010 to October 2014 and the Leader of the Wireless Signal Processing (WSP)

Program at the National ICT Australia (NICTA) from November 2005 to June 2007. His research interests

are in the areas of spatial audio and acoustic signal processing, and multi-channel signal processing. He has supervised over

30 PhD students and coauthored over 200 peer reviewed papers. Professor Abhayapala is an Associate Editor of IEEE/ACM

Transactions on Audio, Speech, and Language Processing. He is a Member of the Audio and Acoustic Signal Processing

Technical Committee (2011-2016) of the IEEE Signal Processing Society. He is a Fellow of the Engineers Australia (IEAust).


Documents

Recent Advances in Active Noise Control Inside Automobile ...€¦ · In an attempt to resolve the above problem, Active Noise Control (ANC) methods were developed where secondary